Latest Articles for Data Science

A new method for improving reward design in reinforcement learning using video signals.

Dynosaur automates instruction-tuning data creation, offering quality and cost benefits.

This study investigates how FOMC communication influences financial markets and monetary policy.

New methods improve finetuning of large models using less memory.

A new method enhances accuracy in estimating health treatment effects.

WebIE tackles the challenges of extracting structured data from web content.

This study evaluates how LLMs enhance training data for commonsense reasoning across languages.

This article discusses enhancing few-shot learning through effective example selection.

New methods improve machine reasoning and explanation without labeled data.

A study on the effectiveness of multilingual LLMs in handling code-switching.

Research focuses on stabilizing solutions using oversmoothing penalties in Banach spaces.

Combining language models and Wikidata improves accurate question answering.

A new method improves insights into gene interactions and spatial dependencies in tissues.

New methods for selecting examples enhance performance in language models.

A new method enhancing AI's decision-making transparency without sacrificing accuracy.

IfQA provides over 3,800 counterfactual questions to challenge open-domain question answering.

A method to identify errors in machine learning models using attention sets.

A new model enhances semi-supervised classification in GNNs through optimized node-wise diffusion.

Researchers enhance language models using sparse networks and Mixture-of-Experts for better performance.

MasakhaPOS provides a dataset for better language processing in Africa.

A method to interpret AI decisions using topology and geometry.

New methods enhance training efficiency and model performance using IKL loss.

A new approach uses clustering indices for faster model selection in data science.

Exploring how block unitriangular matrices aid in machine learning.

A new dataset enhances language models' ability to retain information over time.

MIANet improves segmentation accuracy with limited training data through innovative information integration.

Examining bias in language models and methods to reduce it.

A new method enhances deep learning reliability through better uncertainty assessment.

Discover how multi-task learning improves user recommendations across various platforms.

A new model uses data and physics to predict how particles behave in fluids.

New model improves language identification, focusing on low-resource languages and quality data.

ReXMiner improves information extraction from complex web pages using innovative methods.

A look at how neural networks can approximate solutions to complex equations.

New training method improves language model performance using better example selection.

Using reinforcement learning to improve long-term user engagement in advertising.

A fresh method uses synthetic feedback for better AI alignment with human values.

New method simplifies item suggestions using user conversations.

A method to enhance text classification by reducing the need for labeled data.

A new method enhances the quality of text summarization, especially for Vietnamese content.

New calibration method enhances reliability of deep learning model predictions.